From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933161Ab2GMMiU (ORCPT ); Fri, 13 Jul 2012 08:38:20 -0400 Received: from cantor2.suse.de ([195.135.220.15]:47771 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933135Ab2GMMiS (ORCPT ); Fri, 13 Jul 2012 08:38:18 -0400 Date: Fri, 13 Jul 2012 14:38:17 +0200 From: Jan Kara To: Thomas Gleixner Cc: Jan Kara , Jeff Moyer , LKML , linux-fsdevel@vger.kernel.org, Tejun Heo , Jens Axboe , mgalbraith@suse.com Subject: Re: Deadlocks due to per-process plugging Message-ID: <20120713123817.GC20361@quack.suse.cz> References: <20120711133735.GA8122@quack.suse.cz> <20120711201601.GB9779@quack.suse.cz> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.20 (2009-06-14) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu 12-07-12 00:12:44, Thomas Gleixner wrote: > On Wed, 11 Jul 2012, Jan Kara wrote: > > On Wed 11-07-12 12:05:51, Jeff Moyer wrote: > > > This eventually ends in a call to blk_run_queue_async(q) after > > > submitting the I/O from the plug list. Right? So is the question > > > really why doesn't the kblockd workqueue get scheduled? > > Ah, I didn't know this. Thanks for the hint. So in the kdump I have I can > > see requests queued in tsk->plug despite the process is sleeping in > > TASK_UNINTERRUPTIBLE state. So the only way how unplug could have been > > omitted is if tsk_is_pi_blocked() was true. Rummaging through the dump... > > indeed task has pi_blocked_on = 0xffff8802717d79c8. The dump is from an -rt > > kernel (I just didn't originally thought that makes any difference) so > > actually any mutex is rtmutex and thus tsk_is_pi_blocked() is true whenever > > we are sleeping on a mutex. So this seems like a bug in rtmutex code. > > Thomas, you seemed to have added that condition... Any idea how to avoid > > the deadlock? > > Mike has sent out a fix related to the plug stuff, which I just posted > for the rt stable series. Can you verify against that ? Yeah, that fix from Mike makes us unable to reproduce the problem. But frankly it is a hack and I wouldn't bet a penny there isn't another similar problem hiding elsewhere in the code. Just it would need a different timing / load to trigger. So I think a better solution needs to be found (an advice from JBD maintainer TM ;). Honza -- Jan Kara SUSE Labs, CR